How AI understands
search intent
When someone asks ChatGPT "Which accountant is best for a sole trader?", the platform does not just process the words. It understands the intention behind them. The user is not looking for a definition of accounting. They want a recommendation, a comparison or a shortlist. AI classifies search intent with 80 to 92% accuracy and tailors its answer accordingly. For the majority of business-related queries in ChatGPT, the intent is commercial: comparing, considering, selecting. This article explains how AI analyses the intent behind every question and why this directly determines which businesses get recommended to UK consumers.
92%
accuracy of AI intent classification
73%
of ChatGPT business queries have commercial intent
4.8x
higher selection chance for pages with 15+ recognised entities
54%
of UK adults now use AI tools regularly
What is search intent and why does it matter for AI?
Search intent is the reason behind a query. When someone types "best accountant Manchester" into Google, the intent is clear: they want to find and hire an accountant. When someone asks ChatGPT "What should I look for when choosing an accountant?", the intent is different: they want guidance before making a decision. Same topic, different needs. The type of content that serves each intent is fundamentally different.
Google has always tried to understand intent, but it works primarily through keyword matching and user behaviour signals. If most people who search "conveyancing solicitor" click on comparison pages rather than definitions, Google learns that the intent is commercial and adjusts its results accordingly. That process is indirect and slow.
AI analyses meaning directly
AI search engines take a fundamentally different approach. They do not infer intent from user behaviour. They analyse the meaning of the question directly using natural language processing. ChatGPT reads "Which accountant is best for a sole trader?" and immediately classifies this as a commercial query with local potential. It knows the user wants specific recommendations, not a definition of what sole traders are. That classification happens in milliseconds and determines everything about the answer: which sources are selected, how information is presented and which businesses get mentioned.
For UK businesses, this shift matters enormously. In the Google world, you could rank for a keyword regardless of whether your content matched the user's actual intent. A glossary page defining "conveyancing" could rank for "conveyancing solicitor" if it had enough backlinks. In the AI world, that does not happen. If the intent is commercial (the user wants to hire a conveyancing solicitor), AI will not cite a page that merely defines the term. It will cite pages that compare solicitors, list prices or recommend specific firms.
The consequence is that your content strategy needs to be built around intent, not just keywords. Every page on your website should be designed for a specific intent type. A page answering "What is conveyancing?" serves informational intent. A page titled "Best conveyancing solicitors in Leeds" serves commercial intent. A page with your booking form serves transactional intent. Each needs different content, different structure and different depth.
AI does not match keywords. It matches intent. A page optimised for the right keyword but the wrong intent will not be cited.
The four types of search intent AI recognises
1. Informational intent. The user wants to learn something. "What is a limited company?" "How does Making Tax Digital work?" "What is the difference between freehold and leasehold?" AI responds with explanations, definitions and educational content. It cites sources that demonstrate genuine expertise: detailed articles, authoritative guides, government resources. For informational queries, AI rarely recommends specific businesses. It provides knowledge. Your business benefits by being the source of that knowledge. If your website contains the most comprehensive, clearest explanation of how leasehold works, AI platforms are more likely to cite you as an authority when related commercial questions come up later.
2. Commercial intent. The user is researching before a purchase. "Best project management software for small teams" "Which energy supplier is cheapest for a small business?" "Top estate agents in Bristol." This is where AI visibility is most valuable for businesses. The user is actively comparing options and AI provides a curated shortlist. ChatGPT typically names three to five businesses or products, explains the differences and sometimes recommends one. Being on that shortlist is the AI equivalent of a page-one Google ranking. Sources that AI selects for commercial queries tend to be comparison articles, detailed review pages and websites that clearly explain their offering with specifics like pricing, features and target audience.
3. Transactional intent. The user is ready to act. "Book a plumber in Leeds" "Buy standing desk under 300 pounds" "Register a limited company." AI responds with direct links, booking options or step-by-step instructions. Transactional queries are less common in AI search than in Google (people tend to go directly to websites for purchases), but they are growing. When AI does handle transactional queries, it prioritises sources with clear calls to action, pricing information and straightforward booking or purchase processes. A plumber whose website has an online booking form and transparent pricing has an advantage over one with just a phone number.
4. Local intent. The user wants something nearby. "Good Italian restaurant near me" "Emergency locksmith in Glasgow" "Nurseries in Islington." Local intent is particularly important for UK SMEs. AI platforms handle local queries differently from Google. Google shows a map pack with three results. ChatGPT gives a conversational answer naming specific businesses with brief descriptions. Gemini leans heavily on Google Business Profile data. The sources that matter for local AI intent are Google Business Profile, Trustpilot, Yell.com, industry-specific directories (Checkatrade, Bark, the Law Society) and review aggregators. More about local AI visibility in our guide on how AI shows local businesses.
How AI classifies intent differently from Google
Google classifies intent at the keyword level. It has learned over decades that certain keywords correspond to certain intent types. "Buy" indicates transactional intent. "Best" indicates commercial intent. "What is" indicates informational intent. This classification is based on aggregated user behaviour: which results do people click for which queries?
AI search engines classify intent at the semantic level. They do not look at individual keywords. They analyse the complete meaning of the question, including implicit context. Consider the query "plumber Leeds." In Google, that is a simple local query. In ChatGPT, the platform interprets the broader meaning. Is this someone looking for an emergency plumber right now? Or someone planning a bathroom renovation? The phrasing, the time of day, the conversation history and the specificity of the question all influence how AI classifies the intent.
Conversational context changes everything
The biggest difference is conversational context. In Google, each query stands alone. If you search "accountant" and then "Manchester", Google does not connect the two. In ChatGPT, if you ask "What should I look for in a good accountant?" and then "Any recommendations in Manchester?", the platform understands the second question in the context of the first. It knows you want an accountant recommendation in Manchester, not a general Manchester guide. This means your content needs to be written in a way that answers both standalone questions and follow-up questions within a conversation.
AI platforms also recognise mixed intent. A question like "What is the best CRM for a small consultancy and how much does it cost?" contains both commercial intent (which CRM is best?) and transactional intent (what are the prices?). AI responds to both within the same answer, citing sources that address the full scope of the question. A page that only covers features without mentioning pricing might be passed over in favour of a less detailed page that does include both.
Another key difference: AI recognises implicit intent. "My boiler is making a strange noise" does not contain the word "plumber" or "repair", but AI understands this is likely a request for help with a boiler problem. It may respond with diagnostic information, followed by a suggestion to contact a Gas Safe registered engineer. Your content about boiler problems can surface for queries that never mention your specific service, as long as it is contextually relevant. This is a fundamentally different dynamic from Google, where you need to target specific keywords.
AI recognises implicit intent. "My boiler is making a noise" triggers plumber recommendations even though the word "plumber" is never used. Write content that addresses problems, not just services.
Does AI recommend your business for the right queries?
VestVale automatically monitors whether ChatGPT, Gemini, Claude and Google AI mention your business. Discover which queries trigger your recommendations.
How entities and context help AI understand your business
AI does not process text as a string of words. It processes it as a network of entities and relationships. An entity is a recognisable concept: a business name, a location, a service type, a person, a regulation or a product. When ChatGPT reads your website, it identifies entities and maps the relationships between them. "Smith & Partners" is a business entity. "Manchester" is a location entity. "Employment law" is a service entity. "Tribunal representation" is a sub-service entity. The more entities AI can identify on your page and the clearer the relationships between them, the better it understands what your business does and when to recommend it.
Pages with 15 or more recognised entities have a 4.8 times higher chance of being selected by AI platforms. That does not mean you should stuff your page with random terms. It means you should be specific and comprehensive. Instead of writing "We offer legal services in the north of England", write "Smith & Partners provides employment law advice, tribunal representation, settlement agreements and TUPE transfers for businesses in Manchester, Leeds, Sheffield and Liverpool." Every specific term is an entity that AI can recognise and map.
Entity recognition in practice
Consider how this works for a plumber in Birmingham. A vague page saying "We fix all types of plumbing issues" gives AI almost nothing to work with. A detailed page covering "boiler repair, boiler installation, central heating systems, power flushing, radiator replacement, bathroom fitting, emergency call-outs, Gas Safe registered, Worcester Bosch accredited, Vaillant installer, serving Birmingham, Solihull, Edgbaston and Moseley" gives AI a rich map of entities. When someone asks ChatGPT "Who can install a Worcester Bosch boiler in Solihull?", the specific entity match makes your page far more likely to be cited.
Context also works at the page level. AI analyses how information flows across your entire website, not just individual pages. If your homepage mentions you are a firm of solicitors, your about page lists your team's qualifications, your service pages detail each practice area and your blog covers recent legal developments, AI builds a comprehensive picture. That picture is more convincing than a single page trying to say everything at once.
Structured data reinforces entity recognition. Schema markup explicitly tells AI what each piece of information represents. LocalBusiness schema identifies your business name, address and services as structured entities rather than just text on a page. FAQPage schema marks questions and answers as distinct pairs. Author schema connects content to a named expert with credentials. Together, these signals create a machine-readable map of your business that AI can process efficiently. More details in our article on why structured data matters for AI.
The conversation context dimension adds another layer. If a user has been discussing employment law and then asks "What about redundancy procedures?", AI connects this to the employment law context. Your page on redundancy procedures is more likely to surface if your site establishes a clear employment law authority across multiple pages. This cluster approach to content, where related pages reinforce each other, is far more effective for AI visibility than isolated pages targeting individual keywords.
Read more about how these semantic connections work in our article on why semantic content matters for AI.
How to optimise your content for AI intent matching
Map intent types to pages
Audit your existing content and label each page with its primary intent type: informational, commercial, transactional or local. Identify gaps. Many UK business websites have only transactional pages (service descriptions, contact forms) but lack the informational and commercial content that AI uses to establish expertise and make recommendations.
Answer questions directly
For informational intent, answer the question in the first two sentences. Then provide context, examples and depth. AI extracts the direct answer for its response and uses the surrounding content to verify expertise. "How much does conveyancing cost?" should be answered with a price range immediately, not after three paragraphs of introduction.
Create comparison content
For commercial intent, create content that genuinely helps users compare options. Tables, feature comparisons, pricing breakdowns and honest pros and cons. AI prioritises sources that serve the comparing and evaluating intent. A page listing "5 things to consider when choosing a solicitor" with concrete criteria is more likely to be cited than a page that simply says "We are the best solicitor."
Write for problem-based queries
Many AI queries describe a problem rather than naming a service. "My landlord is not returning my deposit" is not a search for "tenant dispute solicitor" but AI connects the two. Write content that addresses common problems your customers face, using their language rather than industry jargon. This captures implicit intent queries.
Include specific entities
Name specific locations you serve, specific services you offer, specific accreditations you hold and specific brands you work with. Each specific term is an entity that AI can match to user queries. "Gas Safe registered, Worcester Bosch accredited, serving Birmingham and Solihull" is far more useful to AI than "Qualified plumber covering the West Midlands."
Build content clusters
Group related content together with clear internal linking. A cluster of pages covering "employment law overview", "unfair dismissal claims", "settlement agreements", "TUPE transfers" and "tribunal representation" creates a topical authority that AI can recognise. The cluster approach serves follow-up queries where AI draws on conversational context.
How each platform handles intent differently
ChatGPT excels at understanding nuanced commercial intent. When someone asks "I run a small cafe in Brighton and need accounting software that handles VAT returns", ChatGPT identifies multiple intent layers: the business type (cafe), the location (Brighton), the need (accounting software), the specific requirement (VAT returns) and the implicit budget constraint (small business). It then recommends software that matches all these criteria. Your content needs to address these specific combinations to be cited.
Google Gemini handles local intent most effectively because it has direct access to Google Business Profile data, Maps listings and local reviews. When someone asks Gemini about a local service, it cross-references your Google Business Profile, your reviews, your listed services and your location data. For UK businesses focused on local customers, an optimised Google Business Profile is essential for Gemini visibility. Gemini is particularly strong at matching service queries to businesses that explicitly list those services in their profile.
Perplexity and Google AI Overviews
Perplexity handles informational and commercial intent with a strong emphasis on current, verifiable sources. It cites its sources transparently, which means it selects content that it can confidently attribute. For commercial queries, Perplexity favours independent reviews, comparison articles and industry analysis over brand-owned content. Earned media (being reviewed or mentioned by independent publications) carries more weight in Perplexity than self-published content.
Google AI Overviews handle intent in a way most similar to traditional Google, but with more emphasis on providing a complete answer. For informational queries, AI Overviews pull from pages that answer the question directly and comprehensively. For commercial queries, they often show a combination of comparison content and specific business information. The key difference from regular Google is that AI Overviews aim to answer the question without requiring a click, which means your content needs to contain the answer itself, not just a promise that the answer is on your page.
Claude takes a particularly careful approach to commercial intent. It tends to avoid making specific recommendations unless it has strong evidence from multiple independent sources. Claude gives more weight to factual accuracy and is more likely to present options with balanced pros and cons rather than a single recommendation. For businesses wanting to be cited by Claude, having independent reviews and third-party mentions is particularly important.
The common thread across all platforms: intent determines which content types get selected. The better your content matches the actual intent behind the query, the higher your chance of being cited. Mismatched intent is the single biggest reason UK businesses fail to appear in AI answers, more so than poor SEO or weak content. A brilliant article answering the wrong question will never be cited. Learn more in our guide on how AI recommends businesses.
How intent matching works in practice for UK businesses
Scenario 1: A solicitor in Edinburgh. A user asks ChatGPT: "I have been unfairly dismissed from my job in Edinburgh. What are my options?" The intent is mixed: informational (what are my legal options?) and implicitly commercial (I probably need a solicitor). ChatGPT first explains the unfair dismissal process in Scotland, then suggests contacting an employment law solicitor. If your firm has a detailed page explaining unfair dismissal in Scottish law with specific timelines, tribunal processes and ACAS early conciliation, you are a strong candidate for citation. If your website only says "We handle employment disputes" without specifics, AI has no reason to select you.
Scenario 2: An e-commerce business. A user asks Gemini: "Best wireless headphones under 100 pounds for running." The intent is clearly commercial with specific constraints (wireless, under 100 pounds, for running). Gemini will cite sources that compare headphones matching these exact criteria. An online retailer with a curated guide titled "Best running headphones under 100 pounds" with comparison tables, sweat resistance ratings and battery life data has a strong chance of citation. A generic headphones category page will not be selected because it does not match the specific intent.
Scenario 3: A local tradesperson. A user asks ChatGPT: "The hot water in my flat stopped working. What should I check before calling a plumber?" The intent starts as informational but has an implicit commercial follow-up. ChatGPT first provides troubleshooting steps (check the boiler pressure, check the thermostat, check the pilot light). If the user then asks "I have tried those and it still does not work. Who can I call in Leeds?", ChatGPT uses the conversational context to recommend plumbers who specialise in boiler repairs in Leeds. A plumber whose website covers common hot water problems with troubleshooting guides, alongside clear information about their services and service area, is positioned for both the informational and the follow-up commercial query.
Scenario 4: A SaaS company. A user asks Perplexity: "How do I track whether ChatGPT mentions my business?" The intent is informational with strong commercial undertones (the user likely wants a tool). Perplexity will cite articles explaining AI visibility monitoring, and if those articles mention specific tools, the tools get exposure. If your SaaS product is mentioned in independent reviews, comparison articles or industry guides about AI monitoring, Perplexity is more likely to include you. Self-published content alone is less effective on Perplexity than on ChatGPT.
The lesson across all scenarios: understand what your customers are actually asking and create content that matches their intent precisely. Not just the topic, but the purpose behind the question. That is how you get cited.
Frequently asked questions
How accurately does AI classify search intent?
Current AI models classify search intent with 80 to 92% accuracy. That is significantly higher than keyword-based approaches. AI can distinguish between "what is conveyancing" (informational) and "best conveyancing solicitor" (commercial) but also handles more nuanced queries like "my buyer's solicitor is being slow, what can I do?" where the intent is both informational and implicitly commercial.
Do I need different content for different AI platforms?
Not entirely different, but differently optimised. The core content can be the same, but emphasis matters. ChatGPT values depth and expertise signals. Perplexity values freshness and independent sources. Gemini values Google Business Profile data for local queries. Google AI Overviews value structured, section-based answers. Create strong content once, then ensure it has the signals each platform prioritises.
What is the most valuable intent type for UK businesses?
Commercial intent is the most directly valuable because it represents users actively considering a purchase. But informational content builds the expertise signals that make your commercial pages more credible. The most effective strategy covers both: informational content establishes authority, commercial content captures recommendations. Local intent is particularly valuable for SMEs serving specific areas.
How does conversational context affect which businesses AI recommends?
Conversational context means AI considers the entire conversation, not just the latest question. If a user has been discussing kitchen renovations and then asks "Who can help?", AI understands they want a kitchen fitter, not a therapist. This means your content should connect related topics clearly. Internal linking between related service pages helps AI see you as a comprehensive source for an entire topic area.
Can AI tell the difference between genuine expertise and keyword stuffing?
Yes. AI models are trained on vast amounts of text and can distinguish between genuinely knowledgeable content and content that merely repeats keywords. Content written by someone with real expertise contains specific details, practical examples, nuanced advice and concrete figures that keyword-stuffed content lacks. AI rewards depth and specificity, not repetition.
How do I find out what intent types my customers are using?
Start by testing manually. Ask ChatGPT, Gemini and Perplexity the questions your customers typically ask. Note what type of content each platform returns. Then use an AI visibility monitoring tool to track which queries trigger mentions of your business and which do not. That reveals intent gaps: query types where your competitors are visible but you are not.
AI understands what your customers mean, not what they type
VestVale automatically monitors how ChatGPT, Gemini, Claude and Google AI respond to queries about your industry. See which intent types trigger your business and where you are missing.
From £19.95/mo excl. VAT. Cancel monthly.